Nano Banana

Verified against Nano Banana / Gemini 3.1 Flash Image · 2026-07-30

Remove an unwanted object from a photo and rebuild what's behind it

A clean-removal edit that erases a specific object or person from an uploaded photo and reconstructs the space behind it using real surrounding context — floor lines, wall patterns, lighting — instead of a blurred patch.

Nano Banana (Gemini 3.1 Flash Image)Gemini app5 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

You are removing a specific unwanted object from an uploaded photo and reconstructing whatever would realistically be behind it, based on the actual surrounding context in the rest of the image.

SOURCE PHOTO
a living-room photo being prepared for a real-estate listing, patterned hardwood floor, plain painted wall.

OBJECT TO REMOVE
a tall pile of moving boxes stacked in the corner near the window. Remove this completely — no ghosting, no faint outline, no partial remnant left behind.

RECONSTRUCTION CONTEXT
the hardwood floor's plank pattern is visible on both sides of the boxes, and the wall behind them is a plain, unpatterned off-white. Use this specifically to determine what should logically fill the space — continue an existing pattern, floor line, wall texture, or background element through the removed area exactly as it would appear if the object had never been there, rather than filling the gap with a generic blurred or smoothed patch that doesn't match anything else in the photo.

LIGHTING AND SHADOW CLEANUP
the boxes are blocking some window light from reaching that corner of the floor — the corner should look brighter once they're removed. If the removed object was casting a shadow onto a nearby surface, or if it was blocking light that would otherwise be falling on the area behind it, correct for that too — the reconstructed area needs to look lit consistently with the rest of the scene, not left with an orphaned shadow that no longer has anything casting it, and not left artificially dark where removing the object should have actually let more light reach that surface.

EDGE QUALITY
Pay close attention to the boundary where the removed object used to be — any texture, pattern, or line that continues through that area (a floorboard seam, a wallpaper pattern, a horizon line) must align correctly on both sides of the removed area, not show a visible seam, warp, or mismatch where the reconstruction meets the untouched parts of the photo.

WHAT MUST NOT CHANGE
Everything else in the frame — everything not part of the removed object or the small area directly behind it — must remain exactly as it was in the original photo: same color, same lighting, same composition, same crop.

OUTPUT USE
the main listing photo for this room on a real-estate platform — this affects how seamless the reconstruction genuinely needs to be; flag if the specific removal requested is complex enough (a large object blocking a highly detailed or patterned background) that a fully seamless result may need one or two rounds of targeted correction rather than a single clean pass.

OUTPUT
One edited image with the object fully removed and the space behind it reconstructed to look like it was always part of the original photograph.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

Requiring the reconstruction to continue an actual visible pattern — a specific floorboard seam, a specific wallpaper repeat — rather than accepting a generic smoothed patch is what separates a genuinely convincing removal from the telltale soft blur that gives away most quick object-removal edits: real backgrounds usually have some structure (a line, a repeat, a texture direction), and a model given the specific pattern to continue through the gap can extend it correctly, while a model given only "remove this and fill the background" defaults to the easiest visually plausible fill, which is very often a texture-less smudge that reads as an edit the moment anyone looks closely at that exact spot. Second, the lighting-and-shadow cleanup instruction targets a failure mode that's specific to removal edits rather than additions: an object being removed was very likely interacting with the scene's light — blocking it, casting a shadow — and simply deleting the object while leaving its shadow or its light-blocking effect in place produces an image with a shadow that has nothing left to explain it, which is a subtle but real inconsistency a careful viewer notices even without being able to say exactly what looks wrong, so correcting for the object's absence has to include correcting for what its absence changes about the light, not just erasing its visible shape. Third, giving the model permission to flag when a removal is complex enough to need iterative correction — a large object obscuring a highly detailed, patterned background — rather than forcing a single confident pass matters because reconstruction difficulty genuinely scales with how much detail was hidden behind the object: removing something in front of a plain wall is a much easier reconstruction than removing something in front of an intricate patterned rug, and treating both as equally one-shot-solvable sets an unrealistic expectation that leads to accepting a subtly wrong result rather than asking for the targeted second pass the harder case actually needs.

Verified against

Nano Banana / Gemini 3.1 Flash Image Gemini 3.1 Flash Image · 2026-07-30

Changelog

  • 2026-07-30 Initial publish, verified against Nano Banana (Gemini 3.1 Flash Image) removing a stack of moving boxes from a listing photo.

Need this built into your business?

If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.

EXPLORE WHAT SCULT BUILDS
Shoot a catalog-ready e-commerce product photo with Nano BananaA full studio-photography brief for Nano Banana (Gemini 3.1 Flash Image) — material, surface, lighting rig, and angle specified separately so the model renders accurate reflections and shadow behavior instead of a generic glossy render.Nano Banana (Gemini 3.1 Flash Image)Gemini app2026-07-20Turn a hex palette into a brand mood-board flat-lay with Nano BananaA composited overhead flat-lay brief that locks every object and surface in the frame to an exact hex palette — the natural next step after generating a palette with the Colour Palette Generator, built for Nano Banana's literal reading of pasted color codes.Nano Banana (Gemini 3.1 Flash Image)Gemini app2026-07-21Make one surgical edit to a photo without touching anything elseA conversational edit instruction for an uploaded photo that isolates exactly one change and explicitly locks everything else — lighting, pose, background, composition — so Nano Banana's edit doesn't quietly redraw the whole frame.Nano Banana (Gemini 3.1 Flash Image)Gemini app2026-07-21Keep one character consistent across a set of different scenesLocks a defined character's face, proportions, and defining features across a run of separately-described scenes in the same Nano Banana conversation, instead of letting each new generation quietly redesign the character.Nano Banana (Gemini 3.1 Flash Image)Gemini app2026-07-22
All Nano Banana prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY